An electromagnetic interference suppression method and system for eVTOL electronic speed controller with data correction

Abstract

The application discloses an eVTOL electric governor electromagnetic interference suppression method and system with data correction, and belongs to the technical field of aviation power electronics. The application collects electric governor operation, electromagnetic interference characteristics and airborne sensitive device interference monitoring parameters under full flight conditions, constructs a parameter paired sample set, completes the building, constraint setting and training verification of the electric governor electromagnetic interference correlation mapping model, establishes a multi-collection end time synchronization mechanism, synchronously collects real-time operation data and identifies flight conditions, outputs the electric governor controllable parameter adjustment amount through model inference, executes adjustment after completing safety verification, corrects the original data of the sensitive device based on the compensation amount output by the model, and completes model closed-loop optimization. The application realizes adaptive suppression of electric governor electromagnetic interference and synchronous correction of disturbed data, adapts to dynamic changes in full flight conditions, ensures stable and safe electric governor power output, and improves electromagnetic interference suppression effect and condition adaptation capability.

Description

Technical Field

[0001] This invention relates to the field of aviation power electronics technology, and in particular to an eVTOL electrically tunable electromagnetic interference suppression method and system with data correction. Background Technology

[0002] Electric vertical takeoff and landing (EVTOL) aircraft, as core transport equipment in the low-altitude economy and urban air mobility sectors, have seen their application in civil aviation, short-haul transportation, and emergency rescue accelerating in recent years. Distributed electric propulsion systems, as the core power systems of these aircraft, directly determine their flight safety and airworthiness compliance through their operational stability and environmental adaptability. Electronic speed controllers (ESCs), as the core power electronic devices controlling the operation of the drive motors in distributed electric propulsion systems, are responsible for the precise regulation of motor speed and torque, and are crucial components for the stable operation of the aircraft's power system. Currently, electromagnetic compatibility (EMC) design for airborne power electronic devices has become a core assessment item for airworthiness certification in the aviation field. The analysis and suppression of ESC electromagnetic interference characteristics in EVTOL aircraft have become a key technological direction in the industry. In existing research and application of related technologies, two major technical paths have been formed for electromagnetic interference suppression of power electronic devices: hardware and software. The hardware path mainly includes filter circuit design, electromagnetic shielding structure optimization, and grounding system specification design, while the software path mainly includes switching frequency optimization, carrier phase adjustment, and switching timing optimization. At the same time, the industry has also carried out electromagnetic interference characteristic analysis and adaptation design work for the power system operation characteristics under the full flight conditions of aircraft. The relevant technical achievements have been continuously applied and iterated in the research and development of various electric aircraft.

[0003] Current electromagnetic interference (EMI) suppression solutions for ESCs (Electronic Speed ​​Controllers) of electric vertical takeoff and landing (EVT) aircraft mostly employ fixed-parameter hardware suppression and open-loop software parameter optimization. These approaches cannot adapt to the dynamic changes in ESC operating status under all flight conditions. They struggle to adaptively and dynamically adjust EMI suppression parameters based on real-time ESC operating data, the disturbance status of airborne sensitive equipment, and changes in flight conditions, thus failing to achieve real-time optimization of EMI suppression effectiveness. Furthermore, most existing solutions focus only on suppressing EMI at its source, lacking a synchronous correction mechanism for the disturbance-affected data collected from airborne sensitive equipment. This prevents accurate compensation of the original data collected by sensitive equipment after EMI occurs, making it difficult to eliminate the adverse effects of existing EMI on the operation of airborne sensitive equipment. Additionally, they fail to establish a complete closed-loop control chain from data acquisition, parameter inference, suppression execution to effect optimization. Furthermore, the existing solution does not establish a quantitative correlation between the ESC operating status, flight conditions, and the degree of disturbance to airborne sensitive equipment. It cannot accurately match the electromagnetic interference suppression requirements under different flight conditions, nor can it ensure the stability and operational safety of the ESC power output throughout the entire electromagnetic interference suppression process. It cannot balance the electromagnetic interference suppression effect with the reliable operation of the aircraft power system. At the same time, it lacks a continuous optimization mechanism based on actual operating results, making it difficult to continuously improve the adaptability and suppression effect of the solution under all flight conditions. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method and system for suppressing electrically tunable electromagnetic interference with data correction in eVTOL.

[0005] The objective of this invention is achieved through the following technical solution:

[0006] A method for suppressing electrically tunable electromagnetic interference (EMI) with data correction for eVTOL is provided, the method comprising the following steps:

[0007] S1. Collect the ESC operating parameters, electromagnetic interference characteristic parameters, and airborne sensitive equipment interference monitoring parameters under all flight conditions, construct a pairing sample set of parameters under all flight conditions, and complete the construction, constraint setting, and training verification of the ESC electromagnetic interference correlation mapping model based on the pairing sample set of parameters under all flight conditions.

[0008] S2. Establish a time synchronization mechanism for each data acquisition terminal to synchronously acquire real-time ESC operation data, real-time airborne sensitive equipment interference monitoring data, and real-time flight status data, and identify real-time flight conditions.

[0009] S3. Input the real-time ESC operating data, real-time airborne sensitive equipment interference monitoring data and real-time flight conditions into the ESC electromagnetic interference correlation mapping model, output the ESC adjustable parameter adjustment amount, and complete the safety verification and execution adjustment of the ESC adjustable parameter adjustment amount;

[0010] S4. Based on the interference distortion compensation amount output by the electronically controlled electromagnetic interference correlation mapping model, the original data collected by the airborne sensitive equipment is corrected. The adjusted electronically controlled operating data and the corrected airborne sensitive equipment data are collected to complete the closed-loop optimization of the electronically controlled electromagnetic interference correlation mapping model.

[0011] Furthermore, step S1 includes the following sub-steps:

[0012] S1.1. Divide the flight conditions into types according to the full flight envelope of eVTOL, cover all operating scenarios within the full flight envelope, collect the ESC operating parameters, electromagnetic interference characteristic parameters and airborne sensitive equipment interference monitoring parameters under each type of flight condition, complete the timestamp alignment and numerical normalization of all collected parameters, perform data dimension filtering and data redundancy removal operations, and output a standardized parameter dataset.

[0013] S1.2. Construct a full flight condition parameter pairing sample set based on the standardized parameter dataset, divide the full flight condition parameter pairing sample set into training set, validation set and test set according to a set ratio, and perform sample set equalization processing operation.

[0014] S1.3. Construct an electrically tunable electromagnetic interference (EMI) correlation mapping model. The EMI correlation mapping model includes an input layer, a feature fusion layer, a multi-objective inference layer, a constraint verification layer, and an output layer connected in series. The input layer has parallel feature branches, including an EMI operation feature branch, an operating condition classification feature branch, and an interference monitoring feature branch. The outputs of all feature branches are connected to the feature fusion layer. The feature fusion layer uses a bidirectional gated recurrent unit module, and the multi-objective inference layer uses a lightweight gradient boosting tree module. Set the multi-objective weighted loss function, maximum number of iterations, initial learning rate, and learning rate decay step size for the EMI correlation mapping model. The multi-objective weighted loss function includes weighted sub-terms corresponding to the electromagnetic interference suppression loss term, the dynamic performance constraint loss term, and the distortion compensation loss term. Train the EMI correlation mapping model based on the training set and the validation set. Verify the accuracy of the EMI correlation mapping model based on the test set and perform online validity verification. Output the EMI correlation mapping model that has completed training and verification.

[0015] Furthermore, step S2 includes the following sub-steps:

[0016] S2.1. Establish a time synchronization mechanism for each acquisition terminal based on a time-sensitive network, complete the time synchronization calibration of the electronically controlled data acquisition terminal, the airborne sensitive equipment data acquisition terminal, and the flight status data acquisition terminal to meet the set accuracy requirements, set the set sampling frequency for each of the electronically controlled data acquisition terminal, the airborne sensitive equipment data acquisition terminal, and the flight status data acquisition terminal, set the sampling trigger synchronization rules for each acquisition terminal, perform the synchronization link stability verification operation, and output a stable and usable synchronization acquisition link;

[0017] S2.2. Based on the synchronous acquisition link, synchronously acquire real-time ESC operation data, real-time airborne sensitive equipment interference monitoring data and real-time flight status data, complete the timestamp alignment of all acquired data, and output a synchronously aligned real-time operation dataset;

[0018] S2.3. Based on the real-time flight status data in the synchronized real-time running dataset, match the operating condition type of all flight conditions, identify the real-time flight conditions, perform the confidence judgment operation of the identification results, and match the weight parameters of the corresponding operating condition's electronically controlled electromagnetic interference association mapping model.

[0019] Furthermore, step S3 includes the following sub-steps:

[0020] S3.1. Input the real-time ESC operation data, real-time airborne sensitive equipment interference monitoring data, and the identified real-time flight conditions from the synchronized real-time operation dataset into the ESC electromagnetic interference correlation mapping model that has been trained and verified, and output the adjustable parameter adjustment amount of the ESC.

[0021] S3.2. Based on the constraints of the electronically controlled electromagnetic interference correlation mapping model, complete the compliance verification of the adjustable parameters of the electronically controlled system and output a clear verification result;

[0022] S3.3. For the adjustable parameters of the ESC that pass the verification, perform gradient smoothing operation on the adjustment amount and send it to the ESC drive unit to complete the adjustment of the corresponding ESC operating parameters. The carrier phase offset is used for ESC carrier phase interleaving control, the switching timing correction is used for active optimization of the turn-on and turn-off timing of the switching devices, and the dead time adjustment is used for adaptive correction of dead time within the safety threshold. For the adjustable parameters of the ESC that fail the verification, trigger the recalculation process of the ESC electromagnetic interference correlation mapping model.

[0023] Furthermore, step S4 includes the following sub-steps:

[0024] S4.1. Obtain the interference distortion compensation amount output by the electrically tunable electromagnetic interference association mapping model after training and verification, perform real-time iterative update operation of the interference distortion compensation amount, and perform frame-by-frame correction on the original airborne sensitive equipment data in the synchronized real-time running dataset based on the interference distortion compensation amount, and output the corrected airborne sensitive equipment data.

[0025] S4.2. Based on the time synchronization mechanism of each acquisition terminal, collect the adjusted real-time ESC operation data and the corrected airborne sensitive equipment data, and calculate the output residual of the ESC electromagnetic interference correlation mapping model;

[0026] S4.3. Based on the calculated output residuals, complete the closed-loop optimization of the electrically controlled electromagnetic interference (EMI) correlation mapping model, update the weight parameters of the EEMI correlation mapping model, and synchronously update the operating weight parameters of the EEMI correlation mapping model for the corresponding operating conditions; collect new sample data that meets the preset operating condition coverage requirements, store the new sample data in the full flight operating condition parameter pairing sample set, and during the set computing power idle period, retrieve the new sample data to perform incremental training priority scheduling operation to complete the incremental training of the EEMI correlation mapping model. During the incremental training process, the inference speed of the fixed EEMI correlation mapping model does not exceed the set threshold.

[0027] Furthermore, in step S1.1, the ESC operating parameters include carrier frequency, carrier phase, switching timing, dead time, bus voltage, three-phase output current, motor speed, motor torque, switching device rate of change, and switching device junction temperature; the electromagnetic interference characteristic parameters include the radiated electromagnetic interference spectrum, peak frequency, peak intensity, and spectral bandwidth, which are obtained through electromagnetic interference time-domain fluctuation characteristic acquisition operations; the airborne sensitive equipment interference monitoring parameters include the navigation equipment carrier-to-noise ratio, positioning error, number of lock-outs, inertial measurement unit data bias, inertial measurement unit data distortion rate, and flight control communication bus bit error rate, which are obtained through sensitive equipment communication stability index monitoring operations.

[0028] Furthermore, in step S2.3, the operating conditions of the entire flight include vertical takeoff, hovering, forward transition, cruise, vertical landing, and emergency maneuvering. Each operating condition covers a set range of load range, bus voltage range, and ambient temperature range. An environmental disturbance factor fusion analysis is performed, and the matching criteria for the operating condition include flight stage, airspeed, altitude, motor torque command, and remaining battery capacity.

[0029] Furthermore, in steps S3.1 and S3.2, the adjustable parameters of the ESC include carrier phase offset, switching timing correction, and dead time adjustment; the constraints of the ESC electromagnetic interference correlation mapping model include motor torque ripple constraints, ESC conversion efficiency constraints, switching device temperature constraints, dead time safety threshold constraints, and bridge arm conduction logic constraints. Dynamic adaptation and adjustment of the constraints are performed, and each constraint is set with a corresponding boundary value.

[0030] Furthermore, in step S4.3, when completing the closed-loop optimization of the electrically tunable electromagnetic interference (EMI) correlation mapping model, new sample data that meets the preset operating condition coverage requirements is collected, a new sample data purification and screening operation is performed, the new sample data is stored in the full flight operating condition parameter pairing sample set, and during the set computing power idle period, the new sample data is retrieved to complete the incremental training of the EMI correlation mapping model, and the weight parameters and running weight parameters of the EMI correlation mapping model are updated.

[0031] A large-scale eVTOL electrically regulated electromagnetic interference adaptive suppression system is provided, comprising a main control module, an electronically regulated execution module, a sensitive equipment monitoring module, and a data interaction module. The main control module is used to complete flight condition identification, electronically regulated electromagnetic interference correlation mapping model inference, parameter compliance verification, interference compensation calculation, and model closed-loop optimization. The electronically regulated execution module is used to collect electronically regulated operation data and receive and execute parameter adjustment commands. The sensitive equipment monitoring module is used to collect interference monitoring data from airborne sensitive equipment and correct the original collected data. The data interaction module is used to realize time synchronization calibration and full-duplex data transmission between modules, and to perform synchronization control and condition confidence assessment operations between modules.

[0032] The beneficial effects of this invention are:

[0033] (1) By constructing a correlation mapping model between the ESC operating parameters, electromagnetic interference characteristics and airborne sensitive equipment monitoring data, and combining the full-process closed-loop control logic, the adaptive suppression of ESC electromagnetic interference and synchronous correction of disturbed data are realized, adapting to the dynamic operation changes of the ESC under all flight conditions.

[0034] (2) By replacing the traditional fixed hardware filtering and shielding scheme with a software-based dynamic parameter adjustment method, the active source suppression of electromagnetic interference can be achieved without increasing the weight of airborne equipment and material costs, thus solving the problem of insufficient adaptability of hardware schemes to operating conditions.

[0035] (3) By setting multi-dimensional safety constraints for ESC operation, the stability of ESC power output and operational safety are ensured throughout the electromagnetic interference suppression process, balancing the interference suppression effect with the reliable operation of the power system. Attached Figure Description

[0036] Figure 1 A flowchart illustrating the steps of an eVTOL electrically tunable electromagnetic interference suppression method with data correction.

[0037] Figure 2 The following is a flowchart illustrating the specific steps of an eVTOL electrically tunable electromagnetic interference suppression method with data correction, provided as an example. Detailed Implementation

[0038] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0039] Example 1

[0040] See Figure 1 This embodiment provides a method for suppressing electrically tunable electromagnetic interference with data correction in eVTOL, which includes the following steps:

[0041] S1. Collect the ESC operating parameters, electromagnetic interference characteristic parameters, and airborne sensitive equipment interference monitoring parameters under all flight conditions, construct a pairing sample set of parameters under all flight conditions, and complete the construction, constraint setting, and training verification of the ESC electromagnetic interference correlation mapping model based on the pairing sample set of parameters under all flight conditions.

[0042] S2. Establish a time synchronization mechanism for each data acquisition terminal to synchronously acquire real-time ESC operation data, real-time airborne sensitive equipment interference monitoring data, and real-time flight status data, and identify real-time flight conditions.

[0043] S3. Input the real-time ESC operating data, real-time airborne sensitive equipment interference monitoring data and real-time flight conditions into the ESC electromagnetic interference correlation mapping model, output the ESC adjustable parameter adjustment amount, and complete the safety verification and execution adjustment of the ESC adjustable parameter adjustment amount;

[0044] S4. Based on the interference distortion compensation amount output by the electronically controlled electromagnetic interference correlation mapping model, the original data collected by the airborne sensitive equipment is corrected. The adjusted electronically controlled operating data and the corrected airborne sensitive equipment data are collected to complete the closed-loop optimization of the electronically controlled electromagnetic interference correlation mapping model.

[0045] In some embodiments, step S1 includes the following sub-steps:

[0046] S1.1. Divide the flight conditions into types according to the full flight envelope of eVTOL, cover all operating scenarios within the full flight envelope, collect the ESC operating parameters, electromagnetic interference characteristic parameters and airborne sensitive equipment interference monitoring parameters under each type of flight condition, complete the timestamp alignment and numerical normalization of all collected parameters, perform data dimension filtering and data redundancy removal operations, and output a standardized parameter dataset.

[0047] S1.2. Construct a full flight condition parameter pairing sample set based on the standardized parameter dataset, divide the full flight condition parameter pairing sample set into training set, validation set and test set according to a set ratio, and perform sample set equalization processing operation.

[0048] S1.3. Construct an electrically tunable electromagnetic interference (EMI) correlation mapping model. The EMI correlation mapping model includes an input layer, a feature fusion layer, a multi-objective inference layer, a constraint verification layer, and an output layer connected in series. The input layer has parallel feature branches, including an EMI operation feature branch, an operating condition classification feature branch, and an interference monitoring feature branch. The outputs of all feature branches are connected to the feature fusion layer. The feature fusion layer uses a bidirectional gated recurrent unit module, and the multi-objective inference layer uses a lightweight gradient boosting tree module. Set the multi-objective weighted loss function, maximum number of iterations, initial learning rate, and learning rate decay step size for the EMI correlation mapping model. The multi-objective weighted loss function includes weighted sub-terms corresponding to the electromagnetic interference suppression loss term, the dynamic performance constraint loss term, and the distortion compensation loss term. Train the EMI correlation mapping model based on the training set and the validation set. Verify the accuracy of the EMI correlation mapping model based on the test set and perform online validity verification. Output the EMI correlation mapping model that has completed training and verification.

[0049] In some embodiments, step S2 includes the following sub-steps:

[0050] S2.1. Establish a time synchronization mechanism for each acquisition terminal based on a time-sensitive network, complete the time synchronization calibration of the electronically controlled data acquisition terminal, the airborne sensitive equipment data acquisition terminal, and the flight status data acquisition terminal to meet the set accuracy requirements, set the set sampling frequency for each of the electronically controlled data acquisition terminal, the airborne sensitive equipment data acquisition terminal, and the flight status data acquisition terminal, set the sampling trigger synchronization rules for each acquisition terminal, perform the synchronization link stability verification operation, and output a stable and usable synchronization acquisition link;

[0051] S2.2. Based on the synchronous acquisition link, synchronously acquire real-time ESC operation data, real-time airborne sensitive equipment interference monitoring data and real-time flight status data, complete the timestamp alignment of all acquired data, and output a synchronously aligned real-time operation dataset;

[0052] S2.3. Based on the real-time flight status data in the synchronized real-time running dataset, match the operating condition type of all flight conditions, identify the real-time flight conditions, perform the confidence judgment operation of the identification results, and match the weight parameters of the corresponding operating condition's electronically controlled electromagnetic interference association mapping model.

[0053] In some embodiments, step S3 includes the following sub-steps:

[0054] S3.1. Input the real-time ESC operation data, real-time airborne sensitive equipment interference monitoring data, and the identified real-time flight conditions from the synchronized real-time operation dataset into the ESC electromagnetic interference correlation mapping model that has been trained and verified, and output the adjustable parameter adjustment amount of the ESC.

[0055] S3.2. Based on the constraints of the electronically controlled electromagnetic interference correlation mapping model, complete the compliance verification of the adjustable parameters of the electronically controlled system and output a clear verification result;

[0056] S3.3. For the adjustable parameters of the ESC that pass the verification, perform gradient smoothing operation on the adjustment amount and send it to the ESC drive unit to complete the adjustment of the corresponding ESC operating parameters. The carrier phase offset is used for ESC carrier phase interleaving control, the switching timing correction is used for active optimization of the turn-on and turn-off timing of the switching devices, and the dead time adjustment is used for adaptive correction of dead time within the safety threshold. For the adjustable parameters of the ESC that fail the verification, trigger the recalculation process of the ESC electromagnetic interference correlation mapping model.

[0057] In some embodiments, step S4 includes the following sub-steps:

[0058] S4.1. Obtain the interference distortion compensation amount output by the electrically tunable electromagnetic interference association mapping model after training and verification, perform real-time iterative update operation of the interference distortion compensation amount, and perform frame-by-frame correction on the original airborne sensitive equipment data in the synchronized real-time running dataset based on the interference distortion compensation amount, and output the corrected airborne sensitive equipment data.

[0059] S4.2. Based on the time synchronization mechanism of each acquisition terminal, collect the adjusted real-time ESC operation data and the corrected airborne sensitive equipment data, and calculate the output residual of the ESC electromagnetic interference correlation mapping model;

[0060] S4.3. Based on the calculated output residuals, complete the closed-loop optimization of the electrically controlled electromagnetic interference (EMI) correlation mapping model, update the weight parameters of the EEMI correlation mapping model, and synchronously update the operating weight parameters of the EEMI correlation mapping model for the corresponding operating conditions; collect new sample data that meets the preset operating condition coverage requirements, store the new sample data in the full flight operating condition parameter pairing sample set, and during the set computing power idle period, retrieve the new sample data to perform incremental training priority scheduling operation to complete the incremental training of the EEMI correlation mapping model. During the incremental training process, the inference speed of the fixed EEMI correlation mapping model does not exceed the set threshold.

[0061] In some embodiments, in step S1.1, the ESC operating parameters include carrier frequency, carrier phase, switching timing, dead time, bus voltage, three-phase output current, motor speed, motor torque, switching device rate of change, and switching device junction temperature; the electromagnetic interference characteristic parameters include the radiated electromagnetic interference spectrum, peak frequency, peak intensity, and spectral bandwidth, which are obtained through electromagnetic interference time-domain fluctuation characteristic acquisition operations; the airborne sensitive equipment interference monitoring parameters include the navigation equipment carrier-to-noise ratio, positioning error, number of lock-outs, inertial measurement unit data bias, inertial measurement unit data distortion rate, and flight control communication bus bit error rate, which are obtained through sensitive equipment communication stability index monitoring operations.

[0062] In some embodiments, in step S2.3, the operating conditions of the full flight conditions include vertical takeoff, hovering, forward transition, cruise, vertical landing and emergency maneuver conditions; each operating condition type covers a set range of load range, bus voltage range and ambient temperature range, and performs environmental disturbance factor fusion analysis operation. The matching basis of the operating condition type includes flight stage, airspeed, altitude, motor torque command and battery remaining capacity.

[0063] In some embodiments, in steps S3.1 and S3.2, the adjustable parameters of the ESC include carrier phase offset, switching timing correction, and dead time adjustment; the constraints of the ESC electromagnetic interference correlation mapping model include motor torque ripple constraints, ESC conversion efficiency constraints, switching device temperature constraints, dead time safety threshold constraints, and bridge arm conduction logic constraints. The constraint dynamic adaptation adjustment operation is performed, and each constraint is set with a corresponding set boundary value.

[0064] In some embodiments, during step S4.3, when the closed-loop optimization of the electrically tunable electromagnetic interference (EMI) correlation mapping model is completed, new sample data that meets the preset operating condition coverage requirements is collected, a new sample data purification and screening operation is performed, the new sample data is stored in the full flight operating condition parameter pairing sample set, and during the set computing power idle period, the new sample data is retrieved to complete the incremental training of the EMI correlation mapping model, and the weight parameters and running weight parameters of the EMI correlation mapping model are updated.

[0065] A large-scale eVTOL electrically regulated electromagnetic interference adaptive suppression system is provided, comprising a main control module, an electronically regulated execution module, a sensitive equipment monitoring module, and a data interaction module. The main control module is used to complete flight condition identification, electronically regulated electromagnetic interference correlation mapping model inference, parameter compliance verification, interference compensation calculation, and model closed-loop optimization. The electronically regulated execution module is used to collect electronically regulated operation data and receive and execute parameter adjustment commands. The sensitive equipment monitoring module is used to collect interference monitoring data from airborne sensitive equipment and correct the original collected data. The data interaction module is used to realize time synchronization calibration and full-duplex data transmission between modules, and to perform synchronization control and condition confidence assessment operations between modules.

[0066] Example 2

[0067] This embodiment provides a specific implementation process for an eVTOL electrically controlled electromagnetic interference suppression method with data correction. The method is based on a quantitative correlation between the operating state of the electronically controlled controller and the degree of disturbance to sensitive airborne equipment, achieving dynamic adaptive suppression of electromagnetic interference and correction of disturbed data. It comprehensively covers the entire process from data acquisition, model building, online control to closed-loop optimization. Figure 2 As shown, the specific implementation process is as follows:

[0068] S1. Sample set construction and model building and training:

[0069] S1.1. Full-condition parameter acquisition and standardization processing:

[0070] First, the entire flight envelope of the eVTOL (Electronic Vertical Take-Off and Landing) aircraft is classified into operating condition types. eVTOL is an electric aircraft capable of vertical take-off and landing and horizontal cruise. The method described in this embodiment is applied to the electronic speed controller (ESC) link of the distributed electric propulsion system of this type of aircraft. The entire flight envelope refers to the set of all flight states under set flight conditions and performance constraints that enable safe flight. In this embodiment, the corresponding operating condition types are obtained based on the entire flight envelope, covering all operating scenarios within the entire flight envelope. After completing the operating condition type classification, ESC operating parameters, electromagnetic interference characteristic parameters, and airborne sensitive equipment interference monitoring parameters are collected for each operating condition type. An ESC is a power electronic device used to control the operating state of the aircraft's drive motor. In this embodiment, electromagnetic interference is suppressed at its source by adjusting the ESC operating parameters. Electromagnetic interference (EMI) refers to electromagnetic radiation and conducted signals generated during the operation of electronic equipment that can affect the normal operation of surrounding electronic equipment. In this embodiment, electromagnetic interference generated by the ESC operation is suppressed. After all parameters are collected, timestamp alignment and numerical normalization are performed on all collected parameters. Data dimension filtering and data redundancy removal operations are performed. The dimensions of the collected parameters are filtered one by one according to preset rules. Duplicate data and invalid redundant data generated during the collection process are removed, and the core data dimensions that can fully characterize the operation of the ESC, electromagnetic interference characteristics and the disturbance status of sensitive equipment are retained. A standardized parameter dataset is output.

[0071] The operating parameters of the ESC include carrier frequency, carrier phase, switching timing, dead time, bus voltage, three-phase output current, motor speed, motor torque, switching device rate of change, and switching device junction temperature. The carrier frequency refers to the repetition frequency of the switching action of the power switching devices in the ESC, and the carrier phase refers to the phase position of the carrier signal within the period. In this embodiment, the electromagnetic interference spectrum is adjusted by adjusting these two types of parameters. The switching timing refers to the order in which the power switching devices are turned on and off, and the dead time is the time interval between the two switching devices on the upper and lower arms of the bridge arm being in the off state, used to avoid bridge arm shoot-through faults. In this embodiment, the electromagnetic interference generated by the switching action is reduced by adjusting these two types of parameters. The bus voltage refers to the DC supply voltage on the input side of the ESC, the three-phase output current refers to the three-phase AC current output by the ESC to the drive motor, and the motor speed and motor torque refer to the operating speed and output torque of the drive motor, used to characterize the power output state of the ESC. The switching device rate of change refers to the rate of change of voltage and current during the switching process, and the switching device junction temperature refers to the operating temperature of the switching device chip. These parameters are used to characterize the operating state of the switching devices and the source characteristics of electromagnetic interference.

[0072] Electromagnetic interference (EMI) characteristic parameters include radiated EMI spectrum, peak frequency, peak intensity, and spectral bandwidth. An EMI time-domain fluctuation characteristic acquisition operation is performed to continuously collect data on the fluctuation state of EMI over time, obtaining complete variation characteristics of EMI in the time domain. The radiated EMI spectrum refers to the intensity distribution of radiated EMI generated by the electronically controlled switch (ECS) at different frequencies. The peak frequency is the frequency point corresponding to the peak EMI intensity. The peak intensity is the EMI intensity corresponding to the peak frequency. The spectral bandwidth refers to the frequency range covered by the EMI energy. These parameters characterize the complete features of EMI generated by the ECS.

[0073] Interference monitoring parameters for airborne sensitive equipment include the carrier-to-noise ratio (CNR) of the navigation equipment, positioning error, number of lock-outs, inertial measurement unit (IMU) data bias, IMU data distortion rate, and flight control communication bus bit error rate. The system performs communication stability monitoring operations on sensitive equipment, continuously monitoring the stability of the communication connection between the airborne sensitive equipment and the flight control system to obtain stable operating parameters of the communication link. The CNR is the ratio of the carrier signal power received by the navigation receiver to the noise power, used to characterize the reception quality of the navigation signal; positioning error is the deviation between the position data output by the navigation equipment and the actual position; and the number of lock-outs is the deviation between the carrier signal power received by the navigation equipment and the actual position. The number of times the receiver fails to lock onto a satellite signal; these parameters characterize the degree to which navigation equipment is affected by electromagnetic interference. The Inertial Measurement Unit (IMU) is a sensor device used to measure the three-axis acceleration and angular velocity of an aircraft. Data bias refers to the static offset of the sensor output, and data distortion rate refers to the ratio of the deviation between the sensor output data and the actual motion state. These parameters characterize the degree to which the IMU is affected by electromagnetic interference. The flight control communication bus bit error rate refers to the proportion of erroneous symbols to the total transmitted symbols during communication between the flight control system and various airborne devices; this parameter characterizes the degree to which the communication link is affected by electromagnetic interference.

[0074] In this embodiment, the full flight conditions are specifically divided into six categories: vertical takeoff, hovering, forward transition, cruise, vertical landing, and emergency maneuver. The parameter collection coverage and sampling requirements for each category are clearly set.

[0075] Specifically, for the vertical takeoff condition, the load coverage range is 20%-100%, the bus voltage coverage range is 80%-120%, and the ambient temperature coverage range is -40℃-70℃, with a sampling frequency requirement of no less than 100kHz; for the hovering condition, the load coverage range is 30%-90%, the bus voltage coverage range is 85%-115%, and the ambient temperature coverage range is -40℃-70℃, with a sampling frequency requirement of no less than 100kHz; for the forward flight transition condition, the load coverage range is 10%-100%, the bus voltage coverage range is 80%-120%, and the ambient temperature coverage range is -40℃-70℃, with a sampling frequency requirement of no less than 100kHz. Hz; For cruise mode, the load coverage range is 20%-80%, the bus voltage coverage range is 90%-110%, the ambient temperature coverage range is -40℃-70℃, and the sampling frequency requirement is not less than 50kHz; For vertical descent mode, the load coverage range is 20%-100%, the bus voltage coverage range is 80%-120%, the ambient temperature coverage range is -40℃-70℃, and the sampling frequency requirement is not less than 100kHz; For emergency maneuver mode, the load coverage range is 0%-100%, the bus voltage coverage range is 75%-125%, the ambient temperature coverage range is -40℃-70℃, and the sampling frequency requirement is not less than 200kHz. During the data acquisition process, for each of the above operating conditions, at least 20 gradient acquisition points were set according to the corresponding load coverage range, bus voltage coverage range, and ambient temperature coverage range. Each acquisition point completed parameter acquisition for 1000 consecutive sampling cycles. Simultaneously, the sampling parameters of the acquisition equipment were configured strictly according to the sampling frequency requirements of the corresponding operating condition to ensure that the acquired data could completely cover all operating boundaries of the corresponding operating condition and fully characterize the ESC operation and electromagnetic interference characteristics under the corresponding operating condition. For vertical takeoff, forward flight transition, vertical landing, and emergency maneuvering operating conditions with wider load coverage and bus voltage coverage ranges, additional dynamic acquisition sequences of load mutations and transient bus voltage fluctuations were added to capture the dynamic changes in electromagnetic interference during operating condition switching, providing full-boundary sample support for subsequent model training. The parameter acquisition coverage range for each operating condition type is shown in Table 1.

[0076] Table 1. Full Flight Condition Types and Parameter Acquisition Coverage

[0077]

[0078] S1.2. Construction of Paired Sample Sets and Dataset Partitioning:

[0079] Based on the standardized parameter dataset output by S1.1, a full-flight-condition parameter pairing sample set is constructed. This sample set consists of a set of sample data formed by pairing electronic control system (ECS) operating parameters, electromagnetic interference characteristic parameters, and airborne sensitive equipment interference monitoring parameters collected at the same time and under the same operating conditions. In this embodiment, by constructing the pairing sample set, the corresponding relationships between the three types of parameters are established, providing a data foundation for subsequent model training. After completing the construction of the pairing sample set, the full-flight-condition parameter pairing sample set is divided into a training set, a validation set, and a test set according to a set ratio. A sample set equalization operation is performed to balance the number of samples under different operating conditions and states, eliminating the problem of uneven distribution of various types of samples in the sample set and ensuring that the proportion of each type of sample in the dataset meets the requirements for model training. The training set is the dataset used for model parameter fitting, the validation set is the dataset used for accuracy verification and hyperparameter adjustment during model training, and the test set is the dataset used for final accuracy verification after model training. The division of these three datasets ensures that the model has sufficient fitting and generalization capabilities.

[0080] In some specific implementations, after the standardized parameter dataset is output, the parameter pairing sample set is constructed according to the timestamp alignment rule. In the specific implementation process, the electronic control parameters, electromagnetic interference characteristic parameters, and airborne sensitive equipment interference monitoring parameters with a deviation of no more than 1μs under the same timestamp are divided into a pairing sample. After removing invalid samples with missing data or values ​​exceeding the reasonable range, the final full flight condition parameter pairing sample set contains 120,000 valid pairing samples.

[0081] After the sample set is constructed, it is divided into a training set, a validation set, and a test set in a fixed ratio of 7:2:1. The training set contains 84,000 samples and is used for fitting and updating the model weight parameters; the validation set contains 24,000 samples and is used for real-time verification of model accuracy and dynamic adjustment of hyperparameters during training; the test set contains 12,000 samples and is used for final verification of the model's generalization ability after training is completed.

[0082] Stratified sampling was used during the sample set partitioning process to ensure that the proportion of samples of each working condition type in the training set, validation set, and test set is consistent with the proportion in the full sample set. This avoids insufficient model fitting ability for some working conditions due to uneven sample distribution. At the same time, 10% of extreme working condition samples were added to the sample set, including samples corresponding to extreme bus voltage fluctuations, load changes, and extreme ambient temperatures, to improve the model's adaptability to extreme flight scenarios.

[0083] S1.3. Construction, training and validation of the association mapping model:

[0084] An electronically controlled electromagnetic interference (ECE) correlation mapping model is constructed. This model is an algorithmic model used to establish the mapping relationship between ECE operating parameters, flight conditions, the disturbance state of airborne sensitive equipment, and ECE adjustable parameters and interference distortion compensation amounts. In this embodiment, this model is used to realize real-time inference of electromagnetic interference suppression parameters and calculation of disturbance data compensation amounts. In this embodiment, the ECE correlation mapping model is set up with five layers connected in series: an input layer, a feature fusion layer, a multi-objective inference layer, a constraint verification layer, and an output layer. The modules, input content, output content, and processing logic of each layer are clearly defined. The system comprises the following layers: Input layer: Three parallel feature branches. Inputs include ESC operating features, operating condition classification features, and interference monitoring features. Outputs are standardized branch feature vectors. Processing logic involves feature branching input and standardization. Feature fusion layer: Bidirectional gated loop unit module. Inputs are standardized feature vectors from each branch. Outputs are temporally correlated fused feature vectors. Processing logic involves bidirectional temporal feature extraction and multi-dimensional fusion. Multi-objective inference layer: Lightweight gradient boosting tree module. Inputs are temporally correlated fused feature vectors. Outputs are initial inference results for ESC parameter adjustment and distortion compensation. Processing logic involves multi-objective fitting and parallel inference. Constraint verification layer: Boundary constraint verification module. Inputs are initial inference results and preset constraint boundaries. Outputs are verified inference results and anomaly indicators for failed verification. Processing logic involves safety constraint compliance verification and anomaly filtering. Output layer: Result branching output module. Inputs are verified inference results. Outputs are final adjustable ESC parameter adjustment and interference distortion compensation. Processing logic involves branched formatted output of inference results. Each layer is sequentially connected in series according to the order of input layer, feature fusion layer, multi-objective inference layer, constraint verification layer, and output layer. The output of the previous layer directly serves as the input of the next layer, completing the complete inference process. During implementation, the three parallel feature branches of the input layer respectively receive the three types of parameters collected above, and output them to the feature fusion layer after completing the feature de-pathing and standardization processing. The feature fusion layer performs bidirectional extraction and fusion of the input temporal features, eliminates the temporal deviation between multi-dimensional features, and outputs a unified dimension fused feature vector to the multi-objective inference layer. The multi-objective inference layer simultaneously completes the prediction of the electronically tunable parameter adjustment amount and the distortion compensation amount, and outputs the initial inference result to the constraint verification layer. The constraint verification layer combines the preset constraint boundaries to complete the compliance verification of the inference result and filters out abnormal results that do not meet safety requirements. Finally, the output layer completes the de-pathing and formatted output of the result, which is output to the subsequent parameter adjustment and data correction stages respectively, ensuring that the inference result can be directly called and executed by the corresponding stage. The structure, input and output, and processing logic of each layer of the electronically tunable electromagnetic interference correlation mapping model are shown in Table 2.

[0085] Table 2. Hierarchy and Functional Parameters of Electrically Adjustable Electromagnetic Interference Association Mapping Model

[0086]

[0087] The input layer employs parallel feature branches, including an ESC operation feature branch, a flight condition classification feature branch, and an interference monitoring feature branch. The outputs of all feature branches are fed into the feature fusion layer. These three feature branches respectively receive features related to ESC operation, flight conditions, and airborne sensitive equipment interference monitoring, enabling parallel input and preliminary processing of different types of features. The feature fusion layer utilizes a bidirectional gated recurrent unit (Bi-GRU) module, a recurrent neural network structure containing update and reset gates. The update gate controls the extent to which previous state information is incorporated into the current state, while the reset gate controls the extent to which previous state information is ignored. This module can simultaneously extract forward and backward dependency features from time-series data, capturing temporal correlations in long-sequence data.

[0088] In this embodiment, the feature fusion layer receives the standardized feature vectors output from the three branches of the input layer, and inputs the feature vectors of 10 consecutive sampling periods into a time sequence into the bidirectional gated recurrent unit module. The module outputs the forward hidden layer features and the backward hidden layer features respectively. After concatenating the two sets of features, a fixed-dimensional fused feature vector is obtained through global average pooling and output to the multi-objective inference layer.

[0089] The multi-objective inference layer uses a lightweight gradient boosting tree module. Lightweight gradient boosting tree (LightGBM) is an ensemble learning algorithm based on a gradient boosting decision tree framework. It iteratively trains multiple weak learners and then weights and sums the results of the weak learners to obtain the final strong learner. It can handle multi-objective regression tasks simultaneously and has good nonlinear fitting ability and anti-overfitting ability.

[0090] In this embodiment, the multi-objective inference layer receives the fused feature vector output by the feature fusion layer and constructs two parallel regression task branches, which correspond to the regression prediction of the electronically adjustable parameter adjustment amount and the regression prediction of the interference distortion compensation amount, respectively. The two branches share the underlying decision tree structure and output the corresponding prediction results to complete the multi-objective parallel inference.

[0091] The constraint verification layer is used to constrain and verify the initial inference results output by the multi-objective inference layer, ensuring that the output parameters meet the set safe operation requirements and avoiding adverse effects of parameter adjustments on the power output performance of the ESC. After completing the hierarchical and modular construction of the model, the training parameters are set by configuring the multi-objective weighted loss function, maximum number of iterations, initial learning rate, and learning rate decay step size of the ESC electromagnetic interference correlation mapping model. The multi-objective weighted loss function is a function used to measure the deviation between the model's inference results and the true values. In this embodiment, the multi-objective weighted loss function includes weighted sub-terms corresponding to the electromagnetic interference suppression loss term, the power performance constraint loss term, and the distortion compensation loss term, which respectively measure the deviation of the electromagnetic interference suppression effect, the degree of compliance with the power performance constraints, and the accuracy of the compensation for disturbed data. By adjusting the weighting coefficients of each sub-term, a balance is achieved between different optimization objectives. After setting the training parameters, the electrically tunable electromagnetic interference (EMI) correlation mapping model is trained based on the training and validation sets. During training, the model's weight parameters are continuously adjusted using the backpropagation algorithm to reduce the output value of the multi-objective weighted loss function until the model converges. After convergence, the accuracy of the EEMI correlation mapping model is verified based on the test set, and an online validity check is performed. The online running status, inference accuracy, and running stability of the model are checked one by one to confirm that the model meets the requirements of actual flight scenarios. The inference accuracy and generalization ability of the model are verified to meet the set requirements. Finally, the EEMI correlation mapping model that has completed training and verification is output.

[0092] In some specific implementations, after completing the hierarchical construction of the electrically tunable electromagnetic interference (EMC) correlation mapping model, the corresponding training parameters are set, and the model is trained and validated. In the specific implementation process, the weighting coefficients of the EMC suppression loss term, dynamic performance constraint loss term, and distortion compensation loss term in the multi-objective weighted loss function are set to 0.4, 0.35, and 0.25, respectively. The weighting coefficients can be dynamically adjusted according to the optimization requirements of different working conditions. The maximum number of iterations for model training is set to 200, the initial learning rate is 0.05, the learning rate decay step size is 20 iterations, the learning rate decay ratio after each iteration is 0.9, the training batch size is 256 samples, and an early stopping mechanism is adopted. When the loss value of the validation set does not decrease for 15 consecutive iterations, the training is terminated early to avoid model overfitting.

[0093] During training, after each complete iteration, the model's inference accuracy is verified using a validation set. The mean absolute error (MAE) between the inference results of the validation set samples and the true values ​​is calculated. When the MAE exceeds a set threshold, the model's hyperparameters are adjusted, and training continues. After training, the model's generalization ability is verified using a test set. After verification, the model is compressed using INT8 quantization. The compressed model has an inference latency of no more than 1ms and consumes no more than 10% of the onboard controller's computing power, meeting the real-time inference requirements of airborne scenarios.

[0094] S2. Synchronous Data Acquisition and Operating Condition Identification:

[0095] S2.1. Establishment of a multi-acquisition-end time synchronization mechanism:

[0096] A time synchronization mechanism based on Time-Sensitive Network (TSN) is established for each acquisition terminal. TSN is a network communication technology based on Ethernet technology with high-precision time synchronization capabilities, which can achieve microsecond-level time synchronization between multiple nodes. In this embodiment, this technology is used to establish a time synchronization mechanism between the electronically controlled data acquisition terminal, the airborne sensitive equipment data acquisition terminal, and the flight status data acquisition terminal to ensure that the data acquired by different acquisition terminals have a consistent time reference.

[0097] After establishing the time synchronization mechanism, the electronically controlled data acquisition terminal, the airborne sensitive equipment data acquisition terminal, and the flight status data acquisition terminal are calibrated to meet the set accuracy requirements, eliminating time deviations between different acquisition terminals. Following the time synchronization calibration, the corresponding sampling frequencies for each of the electronically controlled data acquisition terminal, the airborne sensitive equipment data acquisition terminal, and the flight status data acquisition terminal are set. Simultaneously, sampling trigger synchronization rules for each acquisition terminal are defined, and a synchronization link stability verification operation is performed. The connection status, synchronization accuracy, and operational stability of the time synchronization link are continuously verified, and any anomalies in the synchronization link are identified and corrected promptly to ensure that the sampling actions of each acquisition terminal are triggered synchronously under the same time reference, ultimately outputting a stable and usable synchronization acquisition link.

[0098] S2.2. Real-time synchronous data collection:

[0099] Based on the synchronous acquisition link output by S2.1, real-time ESC operation data, real-time airborne sensitive equipment interference monitoring data, and real-time flight status data are acquired synchronously. During the acquisition process, the sampling frequency and sampling trigger synchronization rules set by S2.1 are strictly followed to ensure that the time base of all acquired data is consistent. After the data acquisition is completed, all acquired data are timestamped to eliminate minor time deviations generated during the acquisition process, and a synchronized real-time operating dataset is output.

[0100] S2.3. Real-time flight condition identification and parameter matching:

[0101] Based on the real-time flight status data within the synchronized and aligned real-time operational dataset output by S2.2, the flight condition types defined in S1.1 are matched, and a confidence level determination operation is performed on the identification results. The accuracy and reliability of each flight condition identification result are judged one by one, and only identification results that meet the confidence requirements are used in subsequent processes. The flight condition types include vertical takeoff, hovering, forward transition, cruise, vertical landing, and emergency maneuver. Among them, vertical takeoff refers to the flight phase in which the aircraft ascends vertically from the ground to a set altitude; hovering refers to the flight phase in which the aircraft maintains a constant position and altitude in the air; and forward transition refers to the flight phase in which the aircraft descends vertically from the ground to a set altitude. The flight phase is divided into several categories: hovering, cruise (where the aircraft maintains a stable horizontal speed and altitude), vertical descent (where the aircraft descends vertically from a set altitude to the ground), and emergency maneuver (where the aircraft makes maneuver adjustments to respond to unforeseen circumstances). Each flight phase covers a set range of load, bus voltage, and ambient temperature. An environmental disturbance factor fusion analysis is performed, incorporating environmental disturbances during flight into the flight phase determination criteria. Accurate matching of flight phase types is achieved by combining these environmental disturbance conditions with flight phase, airspeed, altitude, motor torque command, and remaining battery capacity. After matching, the real-time flight phase is identified, and the corresponding ESC (Electromagnetic Interference) correlation mapping model's weight parameters are matched to provide adaptation parameters for subsequent model inference.

[0102] In some specific implementations, after establishing the synchronous acquisition link, real-time data acquisition and flight condition identification are performed. Specifically, time-sensitive networking (TSN) is used to achieve time synchronization calibration of the ESC data acquisition terminal, the airborne sensitive equipment data acquisition terminal, and the flight status data acquisition terminal. The time synchronization accuracy after calibration does not exceed 0.5 μs. The sampling frequency of the ESC data acquisition terminal is set to 100 kHz, the sampling frequency of the airborne sensitive equipment data acquisition terminal is 1 kHz, and the sampling frequency of the flight status data acquisition terminal is 100 Hz. The sampling trigger signals of the three acquisition terminals are synchronously issued based on the same time base, ensuring complete alignment of the timestamps of the acquired data. After completing real-time data acquisition, real-time flight condition identification is performed based on the acquired flight status data. A pre-trained lightweight classifier is used for condition identification. The classifier's input consists of five features: flight stage, airspeed, altitude, motor torque command, and remaining battery capacity. The output is the corresponding condition type. The latency for condition identification does not exceed 10 ms, and the identification accuracy is not less than 99%. After completing the operating condition identification, the model operation weight parameters corresponding to the operating condition are matched. For vertical takeoff, vertical landing, and emergency maneuver conditions, the weighting coefficient of the power performance constraint loss term is increased. For cruise conditions, the weighting coefficient of the electromagnetic interference suppression loss term and the distortion compensation loss term is increased, so as to achieve dynamic adaptation of the optimization target under different operating conditions.

[0103] S3. Model Inference and Parameter Tuning Execution:

[0104] S3.1. Model Inference and Parameter Adjustment Output:

[0105] The real-time ESC operation data and real-time airborne sensitive equipment interference monitoring data from the synchronized and aligned real-time operation dataset output by S2.2, and the real-time flight conditions identified by S2.3, are input into the ESC electromagnetic interference correlation mapping model output by S1.3, which has completed training and verification. The model receives corresponding types of input data through three feature branches of the input layer. After feature fusion by the feature fusion layer, multi-objective inference by the multi-objective inference layer, and preliminary constraint verification by the constraint verification layer, the model outputs the adjustable parameters of the ESC. The adjustable parameters of the ESC include carrier phase offset, switching timing correction, and dead time adjustment, which correspond to the adjustment values ​​of carrier phase, switching timing, and dead time in the ESC operation parameters, respectively.

[0106] S3.2. Compliance verification of parameter adjustment amount:

[0107] Based on the constraints of the ESC electromagnetic interference correlation mapping model set in S1.3, a dynamic adaptation adjustment operation of the constraints is performed. According to the real-time flight conditions and ESC operating status, the boundary values ​​of the constraints are dynamically adjusted to adapt the constraints to the operating requirements of the current flight scenario. This completes the compliance verification of the adjustable parameters of the ESC output in S3.1. The constraints of the ESC electromagnetic interference correlation mapping model include motor torque ripple constraints, ESC conversion efficiency constraints, switching device temperature constraints, dead time safety threshold constraints, and bridge arm conduction logic constraints. Each constraint has a corresponding set boundary value. Among them, the motor torque ripple constraint is used to limit the fluctuation range of motor output torque caused by parameter adjustment; the ESC conversion efficiency constraint is used to limit the decrease in ESC energy conversion efficiency caused by parameter adjustment; the switching device temperature constraint is used to limit the operating temperature of the switching device from exceeding the safe range; the dead time safety threshold constraint is used to ensure that the adjusted dead time is not lower than the minimum safe value to avoid bridge arm shoot-through faults; and the bridge arm conduction logic constraint is used to ensure that the conduction and turn-off logic of the switching device meets the safe operation requirements of the ESC.

[0108] In this embodiment, the adjustment step size, adjustment threshold and associated constraint conditions of the electronically adjustable parameters are clearly set, specifically including 5 types of core adjustable parameters. Specifically, the adjustment step size for carrier phase offset is 1°, the minimum adjustment threshold is 0°, and the maximum adjustment threshold is 360°, with the associated constraint being motor torque ripple constraint; the adjustment step size for switch timing correction is 100ns, the minimum adjustment threshold is -2μs, and the maximum adjustment threshold is 2μs, with the associated constraint being switch device temperature constraint; the adjustment step size for dead time adjustment is 50ns, the minimum adjustment threshold is 200ns, and the maximum adjustment threshold is 2μs, with the associated constraint being dead time safety threshold constraint; the adjustment step size for carrier frequency spread spectrum range is 1kHz, the minimum adjustment threshold is -5% of the rated value, and the maximum adjustment threshold is +5% of the rated value, with the associated constraint being ESC conversion efficiency constraint; and the adjustment step size for switch conduction slope adjustment is 5V / μs, the minimum adjustment threshold is 10V / μs, and the maximum adjustment threshold is 100V / μs, with the associated constraint being bridge arm conduction logic constraint. During implementation, the verification process strictly follows the above parameter configuration. First, for each adjustable parameter output, the adjustment step size is checked to ensure it matches the corresponding parameter's adjustment step size, guaranteeing that the adjustment amount can be accurately executed by the ESC drive unit. Second, the adjustment amount is checked to ensure it falls between the corresponding minimum and maximum adjustment thresholds, preventing parameter adjustments from exceeding the ESC hardware's execution range. Finally, the adjustment amount is checked to ensure it meets the set boundary values ​​of the corresponding associated constraints, ensuring that parameter adjustments will not adversely affect the ESC's power output performance or operational safety.

[0109] The verification is considered successful if all check items meet the requirements, and unsuccessful if any item fails. A clear verification result is output after verification. The adjustment step size and adjustment threshold set above also serve as constraints during model training, embedded in the model's loss function to ensure that the adjustment amount of the model output always falls within a safe and executable range. The adjustment range and constraint boundaries of the electrically adjustable parameters are shown in Table 3.

[0110] Table 3 Adjustable Parameter Range and Constraint Boundaries of ESC

[0111]

[0112] S3.3. Parameter Adjustment Execution and Exception Handling:

[0113] The verification result of S3.2 output indicates the controllable parameter adjustment amount of the ESC. A gradient smoothing operation is then performed on the adjustment amount, smoothly transitioning the parameter adjustment amount according to a preset gradient to avoid sudden changes in parameter adjustment affecting the ESC's operating state. This result is then sent to the ESC drive unit to complete the adjustment of the corresponding ESC operating parameters. The carrier phase offset is used for ESC carrier phase interleaving control. Carrier phase interleaving control refers to adjusting the carrier signal phase of each ESC in a distributed multi-ESC system so that the electromagnetic interference peaks generated by the switching actions of each ESC cancel each other out in the frequency domain, reducing the overall electromagnetic radiation intensity of the system. This applies to a distributed system containing six ESCs. The electric propulsion system can sequentially shift the carrier phase of the six electronically controlled switches (ECS) by 60°, causing the switching noise peaks of each ECS to cancel each other out at the same frequency, thus reducing the overall electromagnetic interference (EMI) peak intensity. The switching timing correction is used for active optimization of the turn-on and turn-off timing of the switching devices. By adjusting the turn-on and turn-off times of the switching devices, zero-voltage or zero-current switching states are achieved, reducing the rate of voltage and current change during switching operations and minimizing EMI generation at its source. The dead-time adjustment is used for adaptive correction of the dead-time within a safety threshold, optimizing the dead-time while ensuring safety, and reducing EMI generated by the reverse recovery current of the switching devices. For ECS adjustable parameter adjustments that fail the verification result of S3.2, a recalculation process of the ECS EMI correlation mapping model is triggered. The model re-infers and outputs new ECS adjustable parameter adjustments until the verification result passes.

[0114] S4. Data Correction and Model Closed-Loop Optimization:

[0115] S4.1. Correction of data from airborne sensitive equipment that has been disturbed:

[0116] Obtain the interference distortion compensation amount output by the electrically tunable electromagnetic interference (EMI) association mapping model that has completed training and verification, as output by S1.3. Perform a real-time iterative update operation on the interference distortion compensation amount. Based on the real-time EMI status and the disturbance status of sensitive equipment, continuously iterate and update the interference distortion compensation amount to ensure that the compensation amount matches the current interference status. Based on the interference distortion compensation amount, perform frame-by-frame correction on the original airborne sensitive equipment data in the synchronized and aligned real-time running dataset output by S2.2. Frame-by-frame correction means compensating and canceling the interference distortion component corresponding to each frame of acquired data according to the data acquisition time sequence. In specific implementation, subtract the interference distortion compensation amount of the corresponding frame from each frame of original acquired data to obtain the corrected data after eliminating the influence of EMI. During the correction process, keep the time sequence of the data unchanged to avoid changes in the data time sequence relationship caused by the correction operation. Finally, output the corrected airborne sensitive equipment data.

[0117] S4.2. Model Output Residual Calculation:

[0118] Based on the time synchronization mechanism established in S2.1, the system collects real-time ESC operating data adjusted in S3.3 and corrected airborne sensitive equipment data output in S4.1, maintaining consistency of the time base throughout the acquisition process. After data acquisition, the output residual of the ESC electromagnetic interference correlation mapping model is calculated. The output residual refers to the deviation between the expected optimization effect of the model inference and the actual optimization effect, including deviations in electromagnetic interference suppression effect, dynamic performance constraints, and data correction accuracy. The output residual is used to characterize the model's inference accuracy and optimization effect, providing a basis for subsequent model closed-loop optimization.

[0119] S4.3. Model Closed-Loop Optimization and Incremental Training:

[0120] The closed-loop optimization of the electrically tunable electromagnetic interference (EMI) correlation mapping model is completed based on the output residuals calculated by S4.2. The weight parameters of the model are adjusted by the backpropagation algorithm to reduce the output residuals of the model and improve the inference accuracy of the model. At the same time, the weight parameters of the EMI correlation mapping model for the corresponding working conditions are updated synchronously to ensure the model's adaptability to different working conditions. While completing closed-loop optimization, new sample data that meets the preset operating condition coverage requirements is collected. A new sample data purification and screening operation is performed, cleaning and screening the newly collected sample data to remove abnormal and invalid samples, retaining valid new samples that meet the model training requirements. The new sample data is stored in the full flight operating condition parameter pairing sample set to expand the coverage of the sample set. During the set idle computing power period, the new sample data is retrieved and incremental training priority scheduling is performed. The execution priority of incremental training is scheduled according to the system computing power occupancy status and flight operation status to ensure that incremental training does not affect the core flight control functions. Incremental training of the electrically tunable electromagnetic interference (EMI) correlation mapping model is completed. Incremental training refers to fine-tuning the model based on new sample data without changing the original trained weight parameters, improving the model's adaptability to new operating conditions and scenarios. During incremental training, the inference speed of the EMI correlation mapping model is kept within a set threshold to ensure the real-time performance of the model's online operation.

[0121] In some specific implementations, after calculating the model output residuals, closed-loop optimization and incremental training of the model are performed. Specifically, based on the output residuals, a stochastic gradient descent algorithm is used to fine-tune the model weight parameters online. During fine-tuning, only the weight parameters of the model output layer are updated, without changing the feature extraction weights of the underlying layers. The number of iterations for a single fine-tuning session does not exceed 5, and the time consumed by a single fine-tuning session does not exceed 2ms, ensuring that the online fine-tuning process does not affect the model's real-time inference. Simultaneously, new sample data that meets the preset operating condition coverage requirements is collected during flight. This new sample data includes samples where the model output residuals exceed a set threshold and samples corresponding to operating conditions not covered in the initial sample set. The number of new samples collected during a single flight does not exceed 1000 sets. The collected new sample data is stored in the full flight operating condition parameter pairing sample set, completing the expansion of the sample set. During the idle computing power period when the aircraft is cruising in level flight, newly collected sample data is retrieved to complete the incremental training of the model. During the incremental training, the weight parameters of the bottom feature extraction layer of the model are fixed, and only the weights of the top inference layer are fine-tuned. The number of iterations of incremental training is set to 50, and the initial learning rate is set to 0.01. After training, the inference accuracy and inference speed of the model are verified to ensure that the inference speed does not exceed the set threshold. The updated weight parameters are written to the storage unit of the airborne controller to complete the online update of the model, continuously improving the model's adaptability to all working conditions and the electromagnetic interference suppression effect.

[0122] The method described in this embodiment is executed by a corresponding large-scale eVTOL electrically tunable electromagnetic interference adaptive suppression system. This system includes a main control module, an electronically tunable execution module, a sensitive equipment monitoring module, and a data interaction module. The main control module is responsible for flight condition identification, electronically tunable electromagnetic interference correlation mapping model inference, parameter compliance verification, interference compensation calculation, and model closed-loop optimization, coordinating the entire process's logical execution. The electronically tunable execution module collects electronically tunable operation data, receives parameter adjustment commands from the main control module, and executes corresponding parameter adjustment actions to control the electronically tunable operation status. The sensitive equipment monitoring module collects interference monitoring data from airborne sensitive equipment, receives interference distortion compensation from the main control module, corrects the original collected data, and outputs the corrected sensitive equipment data. The data interaction module enables time synchronization calibration and full-duplex data transmission between modules, performs synchronization control and condition confidence assessment between modules, manages the time synchronization status between modules, and continuously evaluates the confidence of the condition identification results to ensure that data interaction between modules has a consistent time base and a stable communication link.

[0123] This embodiment achieves adaptive suppression of electromagnetic interference (EMI) and synchronous correction of disturbed data for eVTOL electronically controlled aircraft (eVTOL) through a complete technical solution. It can adapt to changes in the eVTOL's operating status under all flight conditions, exhibiting better adaptability compared to fixed-parameter suppression schemes. By establishing a correlation mapping model between eVTOL operating parameters and the disturbed state of airborne sensitive equipment, active suppression of EMI can be achieved at its source, while simultaneously compensating and correcting the generated disturbed data, forming a complete closed-loop control link and reducing the impact of EMI on the normal operation of airborne sensitive equipment. By setting corresponding safety constraints during parameter adjustment, the stability and safety of the eVTOL's power output can be ensured while achieving EMI suppression, avoiding adverse effects on the aircraft's power performance. This embodiment requires no additional hardware filtering and shielding devices, reducing the weight and cost of airborne equipment. Furthermore, through closed-loop optimization and incremental training of the model, the adaptability and suppression effect of the solution can be continuously improved, demonstrating significant engineering application value.

[0124] The above description is merely a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and environments, and can be altered within the scope of the concept described herein through the above teachings or related technologies or knowledge. Modifications and variations made by those skilled in the art that do not depart from the spirit and scope of the present invention should be within the protection scope of the appended claims.

Claims

1. A method for suppressing electrically tunable electromagnetic interference in eVTOL systems with data correction, characterized in that, Includes the following steps: S1. Collect the ESC operating parameters, electromagnetic interference characteristic parameters, and airborne sensitive equipment interference monitoring parameters under all flight conditions, construct a pairing sample set of parameters under all flight conditions, and complete the construction, constraint setting, and training verification of the ESC electromagnetic interference correlation mapping model based on the pairing sample set of parameters under all flight conditions. S2. Establish a time synchronization mechanism for each data acquisition terminal to synchronously acquire real-time ESC operation data, real-time airborne sensitive equipment interference monitoring data, and real-time flight status data, and identify real-time flight conditions. S3. Input the real-time ESC operating data, real-time airborne sensitive equipment interference monitoring data and real-time flight conditions into the ESC electromagnetic interference correlation mapping model, output the ESC adjustable parameter adjustment amount, and complete the safety verification and execution adjustment of the ESC adjustable parameter adjustment amount; S4. Based on the interference distortion compensation amount output by the electronically controlled electromagnetic interference correlation mapping model, the original data collected by the airborne sensitive equipment is corrected. The adjusted electronically controlled operating data and the corrected airborne sensitive equipment data are collected to complete the closed-loop optimization of the electronically controlled electromagnetic interference correlation mapping model.

2. The method according to claim 1, characterized in that, Step S1 includes the following sub-steps: S1.

1. Divide the flight conditions into types according to the full flight envelope of eVTOL, cover all operating scenarios within the full flight envelope, collect the ESC operating parameters, electromagnetic interference characteristic parameters and airborne sensitive equipment interference monitoring parameters under each type of flight condition, complete the timestamp alignment and numerical normalization of all collected parameters, perform data dimension filtering and data redundancy removal operations, and output a standardized parameter dataset. S1.

2. Construct a full flight condition parameter pairing sample set based on the standardized parameter dataset, divide the full flight condition parameter pairing sample set into training set, validation set and test set according to a set ratio, and perform sample set equalization processing operation. S1.

3. Construct an electronically controlled electromagnetic interference correlation mapping model. The electronically controlled electromagnetic interference correlation mapping model includes an input layer, a feature fusion layer, a multi-objective inference layer, a constraint verification layer, and an output layer connected in series. The input layer is set with parallel feature branches, including an electronically controlled operation feature branch, an operating condition classification feature branch, and an interference monitoring feature branch. The outputs of all feature branches are connected to the feature fusion layer. The feature fusion layer uses a bidirectional gated recurrent unit module, and the multi-objective inference layer uses a lightweight gradient boosting tree module. Training parameters for the electrically tunable electromagnetic interference (EMI) correlation mapping model are set, including a multi-objective weighted loss function, maximum number of iterations, initial learning rate, and learning rate decay step size. The multi-objective weighted loss function includes weighted sub-terms corresponding to EMI suppression loss, dynamic performance constraint loss, and distortion compensation loss. The EMI correlation mapping model is trained using the training and validation sets, and its accuracy is verified using the test set, with online validity checks performed. The resulting EMI correlation mapping model is then output as a completed training and validation model.

3. The method according to claim 1, characterized in that, Step S2 includes the following sub-steps: S2.

1. Establish a time synchronization mechanism for each acquisition terminal based on a time-sensitive network, complete the time synchronization calibration of the electronically controlled data acquisition terminal, the airborne sensitive equipment data acquisition terminal, and the flight status data acquisition terminal to meet the set accuracy requirements, set the set sampling frequency for each of the electronically controlled data acquisition terminal, the airborne sensitive equipment data acquisition terminal, and the flight status data acquisition terminal, set the sampling trigger synchronization rules for each acquisition terminal, perform the synchronization link stability verification operation, and output a stable and usable synchronization acquisition link; S2.

2. Based on the synchronous acquisition link, synchronously acquire real-time ESC operation data, real-time airborne sensitive equipment interference monitoring data and real-time flight status data, complete the timestamp alignment of all acquired data, and output a synchronously aligned real-time operation dataset; S2.

3. Based on the real-time flight status data in the synchronized real-time running dataset, match the operating condition type of all flight conditions, identify the real-time flight conditions, perform the confidence judgment operation of the identification results, and match the weight parameters of the corresponding operating condition's electronically controlled electromagnetic interference association mapping model.

4. The method according to claim 1, characterized in that, Step S3 includes the following sub-steps: S3.

1. Input the real-time ESC operation data, real-time airborne sensitive equipment interference monitoring data, and the identified real-time flight conditions from the synchronized real-time operation dataset into the ESC electromagnetic interference correlation mapping model that has been trained and verified, and output the adjustable parameter adjustment amount of the ESC. S3.

2. Based on the constraints of the electronically controlled electromagnetic interference correlation mapping model, complete the compliance verification of the adjustable parameters of the electronically controlled system and output a clear verification result; S3.

3. For the adjustable parameters of the ESC that pass the verification, perform gradient smoothing operation on the adjustment amount and send it to the ESC drive unit to complete the adjustment of the corresponding ESC operating parameters. The carrier phase offset is used for ESC carrier phase interleaving control, the switching timing correction is used for active optimization of the turn-on and turn-off timing of the switching devices, and the dead time adjustment is used for adaptive correction of dead time within the safety threshold. For the adjustable parameters of the ESC that fail the verification, trigger the recalculation process of the ESC electromagnetic interference correlation mapping model.

5. The method according to claim 1, characterized in that, Step S4 includes the following sub-steps: S4.

1. Obtain the interference distortion compensation amount output by the electrically tunable electromagnetic interference association mapping model after training and verification, perform real-time iterative update operation of the interference distortion compensation amount, and perform frame-by-frame correction on the original airborne sensitive equipment data in the synchronized real-time running dataset based on the interference distortion compensation amount, and output the corrected airborne sensitive equipment data. S4.

2. Based on the time synchronization mechanism of each acquisition terminal, collect the adjusted real-time ESC operation data and the corrected airborne sensitive equipment data, and calculate the output residual of the ESC electromagnetic interference correlation mapping model; S4.

3. Based on the calculated output residuals, complete the closed-loop optimization of the electrically controlled electromagnetic interference (EMI) correlation mapping model, update the weight parameters of the EEMI correlation mapping model, and synchronously update the operating weight parameters of the EEMI correlation mapping model for the corresponding operating conditions; collect new sample data that meets the preset operating condition coverage requirements, store the new sample data in the full flight operating condition parameter pairing sample set, and during the set computing power idle period, retrieve the new sample data to perform incremental training priority scheduling operation to complete the incremental training of the EEMI correlation mapping model. During the incremental training process, the inference speed of the fixed EEMI correlation mapping model does not exceed the set threshold.

6. The method according to claim 2, characterized in that, In step S1.1, the ESC operating parameters include carrier frequency, carrier phase, switching timing, dead time, bus voltage, three-phase output current, motor speed, motor torque, switching device rate of change, and switching device junction temperature; the electromagnetic interference characteristic parameters include the radiated electromagnetic interference spectrum, peak frequency, peak intensity, and spectral bandwidth, which are obtained through electromagnetic interference time-domain fluctuation characteristic acquisition operations. The parameters for monitoring interference with airborne sensitive equipment include the carrier-to-noise ratio of navigation equipment, positioning error, number of lock-offs, zero bias of inertial measurement unit data, distortion rate of inertial measurement unit data, and bit error rate of flight control communication bus, which are obtained through the monitoring operation of communication stability indicators of sensitive equipment.

7. The method according to claim 3, characterized in that, In step S2.3, the operating conditions of the full flight conditions include vertical takeoff, hovering, forward transition, cruise, vertical landing and emergency maneuver conditions. Each operating condition covers the load range, bus voltage range and ambient temperature range within a set range. An environmental disturbance factor fusion analysis is performed. The matching criteria for the operating condition type include flight stage, airspeed, altitude, motor torque command and remaining battery capacity.

8. The method according to claim 4, characterized in that, In steps S3.1 and S3.2, the adjustable parameters of the ESC include carrier phase offset, switching timing correction, and dead time adjustment; the constraints of the ESC electromagnetic interference correlation mapping model include motor torque ripple constraints, ESC conversion efficiency constraints, switching device temperature constraints, dead time safety threshold constraints, and bridge arm conduction logic constraints. Dynamic adaptation and adjustment of the constraints are performed, and each constraint is set with a corresponding boundary value.

9. The method according to claim 5, characterized in that, In step S4.3, when completing the closed-loop optimization of the electrically tunable electromagnetic interference (EMI) correlation mapping model, new sample data that meets the preset operating condition coverage requirements is collected, a new sample data purification and screening operation is performed, the new sample data is stored in the full flight operating condition parameter pairing sample set, and during the set computing power idle period, the new sample data is retrieved to complete the incremental training of the EMI correlation mapping model, and the weight parameters and running weight parameters of the EMI correlation mapping model are updated.

10. A large-scale eVTOL electrically tunable electromagnetic interference adaptive suppression system, used to perform the method as described in any one of claims 1-9, characterized in that, It includes a main control module, an electronic speed controller (ESC) execution module, a sensitive equipment monitoring module, and a data interaction module. The main control module is used to complete flight condition identification, ESC electromagnetic interference correlation mapping model inference, parameter compliance verification, interference compensation calculation, and model closed-loop optimization. The ESC execution module is used to collect ESC operation data and receive and execute parameter adjustment instructions. The sensitive equipment monitoring module is used to collect interference monitoring data of airborne sensitive equipment and correct the original collected data; the data interaction module is used to realize time synchronization calibration and full-duplex data transmission between modules, and to perform inter-module synchronization control and operating condition confidence assessment.

Images